Table 3.
Comparison of different prediction methods on a “high-resolution” 3D data set of 73 sequences
| Prediction results for the 3D data set, 73 sequences | |||||
| Method | Topo | Over | Under | Q3 | Q2 |
| PRODIV-TMHMM | 56 (77%) | 13 | 1 | 82% | 87% |
| THUMBUP | 55 (75%) | 1 | 7 | 72% | 83% |
| PRO-TMHMM | 53 (73%) | 12 | 7 | 79% | 85% |
| HMMTOP2.0_multi | 50 (68%) | 8 | 3 | 78% | 88% |
| HMMTOP2.0 | 44 (60%) | 14 | 5 | 75% | 86% |
| S-TMHMM | 44 (60%) | 13 | 10 | 74% | 85% |
| MEMSAT1.8 | 44 (60%) | 9 | 5 | 71% | 86% |
| PHD_htm2.1_msa | 38 (52%) | 8 | 14 | 71% | 85% |
| TMHMM2.0 | 42 (58%) | 10 | 12 | 76% | 86% |
| TopPred2.0 | 38 (52%) | 15 | 8 | 69% | 86% |
| PHD_htm2.1 | 35 (48%) | 12 | 14 | 70% | 85% |
| ERROR | 5% | — | — | 3% | 1% |
All methods were run as described in Table 1.